Sudhakar Mishra

Mohammad Asif, Sudhakar Mishra, Majithia Tejas Vinodbhai et al.

Emotion recognition using EEG signals is an emerging area of research due to its broad applicability in BCI. Emotional feelings are hard to stimulate in the lab. Emotions do not last long, yet they need enough context to be perceived and felt. However, most EEG-related emotion databases either suffer from emotionally irrelevant details (due to prolonged duration stimulus) or have minimal context doubting the feeling of any emotion using the stimulus. We tried to reduce the impact of this trade-off by designing an experiment in which participants are free to report their emotional feelings simultaneously watching the emotional stimulus. We called these reported emotional feelings "Emotional Events" in our Dataset on Emotion with Naturalistic Stimuli (DENS). We used EEG signals to classify emotional events on different combinations of Valence(V) and Arousal(A) dimensions and compared the results with benchmark datasets of DEAP and SEED. STFT is used for feature extraction and used in the classification model consisting of CNN-LSTM hybrid layers. We achieved significantly higher accuracy with our data compared to DEEP and SEED data. We conclude that having precise information about emotional feelings improves the classification accuracy compared to long-duration EEG signals which might be contaminated by mind-wandering.

Deeraj S K, Sadhana Devarajan, Krishna Mehra et al.

Reinforcement learning from verifiable emotion rewards RLVER has produced language models with strong empathetic performance, evaluated on benchmarks that assume cooperative, honest users. Yet real emotional interactions systematically violate this assumption: users gaslight, escalate, and pressure AI systems for unconditional validation, dynamics that cooperative benchmarks cannot surface. We construct the Adversarial Empathy Benchmark AEB and introduce the Emotional Consistency Score ECS to evaluate empathetic robustness under adversarial conditions. AEB comprises six psychologically grounded adversarial trajectory types with discriminative reward structures that penalize formulaic responses; ECS formally disentangles a model's capacity to track user emotional states from its capacity to improve them. In a controlled experiment across eight scenario-matched conditions (think and no-think conditions on 2 RLVER models, and 2 base models (Qwen 1.5B and 7B) with 480 adversarial dialogues), RLVER-PPO-Think substantially outperforms the same-scale untuned baseline (0.963 vs. 0.761, \(p<0.001, r=0.688\)), with zero dialogue collapses and 47\% higher hidden-intention detection. However, ECS remains nearly flat and is not significantly different for RLVER-PPO-Think versus Base-7B-Think (\(p=0.650\)): RL training improves emotional responsiveness without measurable gains in observable state tracking. We interpret the ECS--FS (Final Score) gap as a behavioral/legibility dissociation inside this simulator family, not as evidence about internal understanding or clinical readiness.

Yash Ingle, Jaival Chauhan, Ankit Yadav et al.

Reinforcement learning with verifiable rewards (RLVR) has become a highly effective method for improving the reasoning abilities of Large Language Models (LLMs). Recent research shows that Negative Sample Reinforcement (NSR) -- which focuses on penalizing incorrect steps rather than simply rewarding correct ones -- can match or even exceed the performance of more complex frameworks like PPO and GRPO across the entire Pass@k spectrum. However, current NSR techniques usually apply a fixed penalty throughout the training process and treat every incorrect response with the same weight. To address these limitations, we propose two extensions to the NSR framework: Adaptive Negative Sample Reinforcement. Rather than using a fixed update rule, A-NSR uses time-dependent scheduling functions. In the initial training phases, the system focuses heavily on correcting errors to stabilize the model. As training continues, it shifts toward more subtle and controlled updates. We also introduce Confidence-Weighted Negative Reinforcement, which operates on the principle that different mistakes carry different levels of importance. CW-NSR assigns specific penalty weights based on the model's normalized sequence likelihood. If the model is highly confident in a wrong path, it receives a larger penalty and for uncertain errors -- where the model is effectively exploring -- are penalized less strictly. Our formal analysis shows how these mechanisms govern token-level updates, allowing the model to leverage prior-guided probability redistribution while providing a natural defense against overfitting. We evaluated these methods on difficult reasoning datasets, including MATH, AIME 2025, and AMC23, using the Qwen2.5-Math-1.5B architecture.

Rudray Dave, Vedang Dubey, Smit Deoghare et al.

Recently, Reinforcement Learning from Verifiable Rewards (RLVR) has been established as a highly effective technique for augmenting the math reasoning skills of Large Language Models (LLMs) based on a single instance. Current state-of-the-art 1-shot RLVR models adopt heuristics for selecting instances, mostly based on historical variance in rewards, which we find to be inherently misleading as a measure of transferability value. In this paper, we propose a Selector-Guided Autonomous Curriculum (SGAC) approach, which employs a learnable selector model on a multi-dimensional feature space consisting of success probability, reward variance, output disagreement (entropy), and semantic difficulty level, instead of the static reward variance heuristic. In our empirical evaluation on pools of candidate problems, we observed that output disagreement, rather than reward variance, is the strongest predictor of reasoning gains in subsequent iterations. Leveraging this finding, we develop an autonomous curriculum algorithm for dynamically siphoning candidate problems from a large pool, ranking them by the learned selector, and running micro-bursts of 1-shot GRPO. Our framework is evaluated using the Hendrycks MATH benchmark, with the Qwen2.5-Math-1.5B model serving as the baseline. Our framework obtains an accuracy of 68.0\% on the hold-out dataset, which is better than the accuracy obtained from the state-of-the-art model, 64.0\%, as well as the 1-shot RLVR checkpoint proposed by Wang et al., which achieved an accuracy of 66.0\%. The results confirm that entropy-based intelligent data curation leads to strict reasoning improvement over static training methods, particularly in severely limited data conditions.

Mohammad Asif, Sudhakar Mishra, Ankush Sonker et al.

This research project aims to tackle the growing mental health challenges in today's digital age. It employs a modified pre-trained BERT model to detect depressive text within social media and users' web browsing data, achieving an impressive 93% test accuracy. Simultaneously, the project aims to incorporate physiological signals from wearable devices, such as smartwatches and EEG sensors, to provide long-term tracking and prognosis of mood disorders and emotional states. This comprehensive approach holds promise for enhancing early detection of depression and advancing overall mental health outcomes.